Structural Characterization of Porcine Adeno-Associated Virus Capsid Protein with Nuclear Trafficking Protein Importin Alpha Reveals a Bipartite Nuclear Localization Signal

Adeno-associated viruses (AAV) are important vectors for gene therapy, and accordingly, many aspects of their cell transduction pathway have been well characterized. However, the specific mechanisms that AAV virions use to enter the host nucleus remain largely unresolved. We therefore aimed to reveal the interactions between the AAV Cap protein and the nuclear transport protein importin alpha (IMPα) at an atomic resolution. Herein we expanded upon our earlier research into the Cap nuclear localization signal (NLS) of a porcine AAV isolate, by examining the influence of upstream basic regions (BRs) towards IMPα binding. Using a high-resolution crystal structure, we identified that the IMPα binding determinants of the porcine AAV Cap comprise a bipartite NLS with an N-terminal BR binding at the minor site of IMPα, and the previously identified NLS motif binding at the major site. Quantitative assays showed a vast difference in binding affinity between the previously determined monopartite NLS, and bipartite NLS described in this study. Our results provide a detailed molecular view of the interaction between AAV capsids and the nuclear import receptor, and support the findings that AAV capsids enter the nucleus by binding the nuclear import adapter IMPα using the classical nuclear localization pathway.


Introduction
Adeno-associated viruses (AAVs) have become an intensively studied group of viruses due to their utility and status as the world's top candidate for use as viral vectors for gene therapy [1]. As members of the family Parvoviridae and genus Dependopavovirus [2,3], these single-stranded DNA viruses are relatively small (~4.7 kilobases) with a simple linear genome encoding two primary genes, Rep and Cap, flanked by inverted terminal repeats [4] that form hairpin structures by self-annealing [5]. There are 13 recognized serotypes [6] and over 150 different variants of AAV, with many displaying specific tissue tropism [7] determined by the receptor binding of the AAV capsid with the cell surface [8]. Despite a high prevalence of infection among vertebrates (the human population has been shown to have seroprevalence ranging from 15-90% depending on the AAV serotype [9][10][11][12]), AAVs are not known to cause physical disease manifestations [13], and possess only a minimal capacity to invoke the immune response [14]. This can be attributed to the inability of AAVs to replicate in host cells without the co-infection with another helper virus [15,16], such as herpesviruses, adenoviruses, or papillomaviruses [11].
Due to their safety and simplicity, AAVs are prominent candidate vectors in the field of gene therapy. Furthermore, AAVs are able to be tailored for a wide range of therapeutic purposes [17]. For example, studies have used sequence comparisons of different AAVs to identify residues which may be responsible for tissue tropism, and engineered these Viruses 2023, 15, 315 2 of 16 residues from one AAV to another, with the resulting chimeras transducing certain tissues more efficiently [18,19]. Currently, there are well over 130 clinical trials underway using AAV vectors [20], and currently three registered and approved drugs for clinical use [21].
Despite the numerous advantages of using AAV vectors for gene therapy, there are a number of issues that need to be considered and/or overcome. One important consideration is the presence of pre-existing immunity within human populations. Although the magnitude of the immune response to AAV is relatively small during natural infection or gene therapy transduction (making the process of therapeutic treatment safe), specific antibodies are still raised that may neutralize the virion [9,10]. This does present an issue when patients display an anamnestic antibody response that may reduce the overall efficacy of an approved AAV vector. Currently, all AAV vectors used in clinical trials have been based on primate-infecting isolates; however, to minimize the drawbacks arising from pre-existing immunity, AAV vectors not naturally encountered within primate populations are being closely studied. Significant differences in capsid sequences likely result in low or no cross-reactivity to human anti-AAV antibodies [22,23]. Reduced neutralization would make a substantial difference in transduction efficiency and subsequent transgene expression, and it has been demonstrated that a porcine-derived AAV was not neutralized by immunoglobulin G from a pool made of over 10,000 human donors [24]. A study published in 2014 by the same group [25] found that not only could different porcine AAV strains transduce mouse tissue, but that they did so with varying degrees of specificity, indicating that a unique tissue tropism profile can also be observed in heterologous host organisms. If porcine AAVs are to be seriously considered as a human gene therapy vector, further understanding of their biological interactions is required, including the mechanism through which porcine AAVs migrate through the cell, and how they interact with human host proteins to achieve this.
All AAVs for which cellular entry has been investigated have been demonstrated to transit the cell via a similar pathway. The capsid attaches to primary and secondary receptors of the cell surface [8,[26][27][28], and the virion is then internalized through the use of multiple endocytic routes [29][30][31]. The translocation of AAV capsids into the host nucleus is a crucial step in this pathway, however, little is known about the interactions between AAV capsids and host transport proteins that allow import to occur [32].
In the classical nuclear import pathway, the cellular transport protein importin-α (IMPα), binds to a cluster of basic residues within a cargo protein [33] known as the NLS [34]. The protein IMPβ then binds to the IBB domain of IMPα [35], and this heterotrimeric complex is then translocated through the nuclear pore complex and into the nucleus. Here, RanGTP then dissociates the complex [36], and IMPα and IMPβ are recycled back into the cytoplasm.
The Cap gene of AAV encodes three capsid proteins that share a C-terminal sequence and combine to form the shell-like outer structure of the virus [37,38]. VP1 is encoded by the entirety of the Cap gene, and as a result is the largest capsid protein containing both N-terminal and C-terminal domains [39,40]. VP2 is translated from an alternatively spliced mRNA, and though it is encoded in the same reading frame as VP1, it is smaller with translation initiating on a start codon~137 residues downstream of the VP1 start site [39,40]. VP3 is similarly expressed from an alternate mRNA through leaky scanning and is the smallest capsid protein, only maintaining the C-terminal domain from residuẽ 200 onwards [39,40]. These three capsid isoforms interlock to form the AAV capsid in a ratio of 1:1:10 (VP1:VP2:VP3), with the N-terminus of VP1 and VP2 located as a flexible segment inside the capsid shell [41]. During endosomal escape, conformational changes occur. In particular, the N-terminal regions of VP1 and VP2 become externally exposed on capsids via extrusion through the virion pores [42,43].
Studies have identified four basic regions (BR) in the AAV VP1 sequence [44]. Basic regions 1 (BR1), 2 (BR2), and 3 (BR3) are located on the N-terminal domain whilst basic region 4 (BR4) is on the C-terminal domain [44]. VP1 contains all four of the basic regions, VP2 retains only BR2, BR3, and BR4, whilst the much shorter VP3 only contains BR4 [44]. Studies have shown that mutation to BR1 altered nuclear transduction [44,45], and mutation of BR2 and BR3 reduced AAV VP1/VP2 to transduce into the nucleus [46]. Interestingly, mutational studies of BR4 have shown this region to have no impact on nuclear import of the capsid protein [44,46], but that this region does contribute to virion assembly [46].
Despite evidence supporting that these BR1-3 act as NLSs, very little data are available to define the way AAV capsids interact with host nuclear import receptors. Similarly, no evidence has yet been reported that could determine whether there is any isotype specificity to be found in interactions between the AAV capsid and different host importin proteins. We recently reported that an AAV porcine strain (Po1) [24] contained a region of the Nterminal capsid domain that is able to act as an NLS [47]. It was revealed that this AAV Po1 capsid NLS bound to mouse IMPα2 (a homolog of human IMPα1), with a high-resolution crystal structure of the complex defining the specific residues directly involved in this interaction. This was the first time a structure had been obtained revealing an importin protein in complex with a suspected AAV capsid NLS.
Comparison of the sequence of AAV Po1 VP1 to primate AAV VP1 proteins revealed a similarity to human AAV5 (a shared amino acid similarity of~85%) ( Figure 1). Intriguingly, for AAV5 and AAV Po1 specifically, the positively charged cluster of amino acids comprising BR2 and BR3 are not congruent with the equivalent regions of the capsid amino acid sequence for the 12 other serotypes. Interestingly, AAV5 and AAV Po1 do not appear to hold a BR2 similar to that of the 12 other serotypes, and do not seem to contain any version of BR3 seen amongst the other serotypes. The region where BR2 would be expected to be observed on both AAV5 and Po1 lacks the prominent positively charged residues that form BR2. Instead, further downstream between where BR2 and BR3 would typically be found, is a region dense with positively charged residues ( Figure 1). This overlapping region found on AAV5 and AAV Po1 is the suspected NLS sequence we have previously reported [47]; however, the homologous BR1 which was observed to be present in all the human AAV serotypes, and AAV Po1 was not investigated in our earlier study. Despite the BR1 sequence residing further upstream from the AAV5 and AAV Po1 BR2 than it does for other AAV serotypes, it remains possible that these two distinct basic regions may cooperate as a bipartite NLS during IMPα interactions.
Studies have identified four basic regions (BR) in the AAV VP1 sequence [44]. Basic regions 1 (BR1), 2 (BR2), and 3 (BR3) are located on the N-terminal domain whilst basic region 4 (BR4) is on the C-terminal domain [44]. VP1 contains all four of the basic regions, VP2 retains only BR2, BR3, and BR4, whilst the much shorter VP3 only contains BR4 [44]. Studies have shown that mutation to BR1 altered nuclear transduction [44,45], and mutation of BR2 and BR3 reduced AAV VP1/VP2 to transduce into the nucleus [46]. Interestingly, mutational studies of BR4 have shown this region to have no impact on nuclear import of the capsid protein [44,46], but that this region does contribute to virion assembly [46].
Despite evidence supporting that these BR1-3 act as NLSs, very little data are available to define the way AAV capsids interact with host nuclear import receptors. Similarly, no evidence has yet been reported that could determine whether there is any isotype specificity to be found in interactions between the AAV capsid and different host importin proteins. We recently reported that an AAV porcine strain (Po1) [24] contained a region of the N-terminal capsid domain that is able to act as an NLS [47]. It was revealed that this AAV Po1 capsid NLS bound to mouse IMPα2 (a homolog of human IMPα1), with a highresolution crystal structure of the complex defining the specific residues directly involved in this interaction. This was the first time a structure had been obtained revealing an importin protein in complex with a suspected AAV capsid NLS.
Comparison of the sequence of AAV Po1 VP1 to primate AAV VP1 proteins revealed a similarity to human AAV5 (a shared amino acid similarity of ~85%) ( Figure 1). Intriguingly, for AAV5 and AAV Po1 specifically, the positively charged cluster of amino acids comprising BR2 and BR3 are not congruent with the equivalent regions of the capsid amino acid sequence for the 12 other serotypes. Interestingly, AAV5 and AAV Po1 do not appear to hold a BR2 similar to that of the 12 other serotypes, and do not seem to contain any version of BR3 seen amongst the other serotypes. The region where BR2 would be expected to be observed on both AAV5 and Po1 lacks the prominent positively charged residues that form BR2. Instead, further downstream between where BR2 and BR3 would typically be found, is a region dense with positively charged residues ( Figure 1). This overlapping region found on AAV5 and AAV Po1 is the suspected NLS sequence we have previously reported [47]; however, the homologous BR1 which was observed to be present in all the human AAV serotypes, and AAV Po1 was not investigated in our earlier study. Despite the BR1 sequence residing further upstream from the AAV5 and AAV Po1 BR2 than it does for other AAV serotypes, it remains possible that these two distinct basic regions may cooperate as a bipartite NLS during IMPα interactions.  Our present study reveals that the upstream BR1 region of the AAV Po1 capsid does indeed contribute to IMPα binding as part of an extensive bipartite NLS. A highresolution structure demonstrates that BR1 (QAKKRVL 125 ) interacts with IMPα at the minor cargo binding site, and the BR2-like NLS region (PKKKKAR 155 ) we previously identified occupies the major site. Biochemical protein-protein interaction assays revealed that the bipartite AAV Po1 capsid NLS binds across all three IMPα subfamilies in preference to IMPβ. Furthermore, quantitative fluorescence polarization assays (FP) demonstrated a far superior binding strength of the bipartite NLS compared to the originally reported monopartite BR2-like region. These data allow insight into the mechanisms of AAV and IMPα binding, opening the potential for a much more direct targeting of capsid engineering. Evidently, understanding that bipartite NLSs are significantly superior in binding importin isoforms compared to monopartite NLSs will aid in the creation of an AAV capsid with an increased efficiency to translocate into a cell nucleus, and subsequently the overall effectiveness of AAV for gene therapy.

Gene Construct Design
The amino acid sequence of AAV Po1 VP1 protein was sourced through GenBank (accession ACN42940.1). A fragment of the VP1 gene corresponding to the amino acid residues 114 GKAVFQAKKRVLEPFGLVEEPVKTAAKGERIDDHYPKKKKARIEE TEAGTSGAQQLQ IPAQP 175 (referred to herein as AAV Po1 Cap-BR), was codon optimized for expression in E. coli and cloned into the pGEX4T-1 vector (GenScript, USA) at BamHI sites with an additional N-terminal tobacco etch virus (TEV) protease site for GST-tag cleavage. Mouse IMPα2 (mImpα2DIBB; His tag, no TEV site) was encoded by a pET30a expression vector as previously detailed [48].

Recombinant Expression and Purification
Expression and co-purification of AAV Po1 VP1 BR protein and IMPα2 was undertaken as described previously [47]. Overexpression of importin proteins α1, α2, α3, α5, α7, and β was performed in E. coli pLysS cells via the auto-induction method [51]. Postinduction, bacterial cells were pelleted via centrifugation at 6000 rpm for 20 min, and were subsequently resuspended using 20 mL per 2 L bacterial culture with HIS buffer A (50 mM phosphate buffer, 300 mM NaCl, 20 mM imidazole, pH 8), and lysed via two freeze-thaw cycles in addition to the application of 1 mL of 20 mg/mL lysozyme (Sigma-Aldrich, St. Louis, MI, USA), and 10 µL of 50 mg/mL DNase (Sigma-Aldrich, USA) at room temperature for 1 h. Soluble extract was harvested via centrifugation at 12,000 rpm for 30 min, and filtered through a 0.45 µm low protein affinity filter. Soluble extract was loaded and injected over a 5 mL HisTrap HP column (GE Healthcare, Chicago, IL, USA), and washed with twenty column volumes of His buffer A on an AKTApurifier FPLC (GE Healthcare, USA). An increasing concentration gradient of imidazole (20 mM to 500 mM) (ChemSupply, Gillman, SA, Australia) eluted the protein off the column, and resulting fractions were pooled. Size-exclusion chromatography was employed to further purify proteins using a HiLoad 26/60 Superdex 200 column (GE Healthcare, USA), pre-equilibrated in GST buffer A (50 mM Tris, 125 mM NaCl). Corresponding fractions of elution volume and protein size were collected and concentrated using Amicon MWCO 10 kDa filter (Merck Millipore, Burlington, MA, USA), and aliquoted for storage at −80 • C. Samples were assessed for purity by SDS-PAGE at 165 V for 30 min on a 4-12% Bis-Tris plus gel (Thermo Fisher Scientific, Waltham, MA, USA).

Crystallization of AAV Po1 Cap-BR and IMPα2 Complex
The protein complex was crystallized using the hanging drop vapour diffusion method with a final crystallization condition of 0.75 M Na Citrate, 0.1 M HEPES pH 6.5 mixed in a 1:1 ratio with a protein complex solution and crystallized at 22 • C. Rod-shaped crystals (100 × 10 × 10 µM) grew after 2 days of incubation. Crystals were collected and

Data Collection and Structure Determination
X-ray diffraction data were obtained from the Australian Synchrotron on the MX2 beamline [52]. Data were indexed and merged through iMOSFLM [53], prior to scaling and subsequent merging via Aimless [54,55]. Phasing was undertaken using molecular replacement in Phaser [56], and the structure 4OIH from the Protein Data Bank (PDB) was used as a search model. Final model rebuilding and relative refining was performed in Coot [57][58][59] and Phenix [60], respectively.

Fluorescence Polarization Assay
Synthetic FITC tagged AAV Po1 capsid NLS peptides 143 RIDDHYPKKKKARIEETEAG TSG 165 (monopartite NLS) and 119 QAKKRVLEPFGLVEEPVKTAAKGERIDDHYPKKKKAR IEE 158 (bipartite NLS; referred to as AAV Po1 Cap-BR) were obtained from GenScript. Peptides were incubated (2 nm) with two-fold serially diluted importin isoform (α1, α2, α3, α5, α7, and β) concentrations (starting concentration 4.5 µM) across 23 wells to a complete volume of 200 µL per well with GST buffer A (50 mM Tris, 125 mM NaCl). Fluorescence polarization measurements were recorded using a CLARIOstar Plus plate reader (BMG Labtech, Germany) in 96-well black Fluotrac microplates (Greiner Bio-One, Austria). Assays were performed in three independent experiments each containing an appropriate negative control (no importin binding partner). Data from the three independent experiments were assessed using GraphPad Prism (Prism 9, Version 9.3.1) and a binding curve fitted to determine the dissociation constant (K D ).

Purification of the IMPα2 and AAV Po1 Cap-BR Complex
Our earlier structural study demonstrated the manner in which the AAV Po1 VP1 suspected monopartite NLS was able to bind with IMPα2, providing the first published data on the structural interactions between an AAV capsid and a host nuclear import protein [47].
Here we used a similar method to gather further data and structural evidence to determine IMPα2 binding in the context of a larger portion of the AAV Po1 VP1 containing residues that span the positions of all three N-terminal BRs. AAV Po1 Cap-BR and IMPα2 were individually expressed within E. coli cells prior to co-purification via Ni-affinity and size exclusion chromatography. Recombinantly expressed protein extracts were combined, and His-IMPα2 and GST-Po1 Cap-BR were observed to interact despite the presence of non-specific proteins within the E. coli lysate [ Figure 2A,B]. Ni-affinity purification of E. coli lysate showed binding between IMPα2 and AAV Po1 Cap-BR in what appears to be a 1:1 ratio [ Figure 2A,B]. The GST-affinity tag was cleaved from AAV Po1 Cap-BR using TEV proteolysis, and then the complex was further purified via size-exclusion chromatography [ Figure 2A,C]. Removal of the remaining contaminating GST affinity tag was performed via injection of the size exclusion-purified sample over a glutathione column with collection and concentration of the column flow-through [ Figure 2A]. Samples were taken at every stage and examined via SDS-PAGE analysis. The complex formed by AAV Po1 Cap-BR and IMPα2 was observed to be stable over multiple purification steps. The purified complex was concentrated to 34 mg/mL for use in crystallization trials.
appears to be a 1:1 ratio [ Figure 2A,B]. The GST-affinity tag was cleaved from AAV Po1 Cap-BR using TEV proteolysis, and then the complex was further purified via size-exclusion chromatography [ Figure 2A,C]. Removal of the remaining contaminating GST affinity tag was performed via injection of the size exclusion-purified sample over a glutathione column with collection and concentration of the column flow-through [ Figure 2A]. Samples were taken at every stage and examined via SDS-PAGE analysis. The complex formed by AAV Po1 Cap-BR and IMPα2 was observed to be stable over multiple purification steps. The purified complex was concentrated to 34 mg/mL for use in crystallization trials.

Protein Crystallization and Data Collection
The IMPα2:AAV Po1 Cap-BR complex was successfully crystallized [ Figure 2D] via the hanging drop vapour diffusion method [61], and short rod-shaped crystals diffracted [ Table 1] on the MX2 beamline at the Australian Synchrotron. The data were indexed and integrated in iMOSFLM [53], prior to merging and scaling performed in Aimless [ Table 1]. Molecular replacement was performed using Phaser to solve the structure using the PDB 4OIH as a search model [62]. Rebuilding and refinement allowed for the structure to be resolved to a resolution of 2.6 Å and refined to a R work /R free of 0.21/0.24 via iterative Cap-BR (residues 120-124; 148-157), and 25 water molecules. The stereochemistry and other refinement statistics are presented in Table 1.

Binding Determinants of the AAV Po1 Cap-BR: IMPα2 Complex
The modelled high resolution IMPα2 structure maintained the expected α-helical conformation of the ten sequential armadillo (ARM) motifs, as previously observed and described [63]. The AAV Po1 Cap-BR was observed to bind within the interaction sites typical of IMPα and NLS complexes, with both the major site of IMPα (ARM 2-4, P1-P5 sites), and minor site (ARM 6-8, P1'-P3' sites) involved in binding interactions (Figure 3). This binding is typical of a bipartite NLS observed in other structures [64][65][66]. It was observed that the overall binding buries a total surface area of 1238.7 Å 2 and is mediated by 16 hydrogen bonds, 12 in the major binding site and 4 in the minor, with a total of 4 salt bridges, 2 in the major binding site and 2 in the minor binding site ( Figure 4B,C) ( Table 2).  . This structure has been deposited to the PDB and given the code: 8FK3.

Binding Determinants of the AAV Po1 Cap-BR: IMPα2 Complex
The modelled high resolution IMPα2 structure maintained the expected α-helical conformation of the ten sequential armadillo (ARM) motifs, as previously observed and described [63]. The AAV Po1 Cap-BR was observed to bind within the interaction sites typical of IMPα and NLS complexes, with both the major site of IMPα (ARM 2-4, P1-P5 sites), and minor site (ARM 6-8, P1′-P3′ sites) involved in binding interactions (Figure 3). This binding is typical of a bipartite NLS observed in other structures [64][65][66]. It was observed that the overall binding buries a total surface area of 1238.7 Å 2 and is mediated by 16 hydrogen bonds, 12 in the major binding site and 4 in the minor, with a total of 4 salt bridges, 2 in the major binding site and 2 in the minor binding site ( Figure 4B,C) ( Table 2). Table 2. Hydrogen bond and salt bridge interactions.    AAV Po1 VP1 residues of the proposed BR2 site occupy the major cargo binding site of IMPα2, in congruence with our prior publication utilizing a shorter AAV Po1 [47]. For the structure obtained with the more extensive AAV Po1 Cap-BR, the residue Lys 151 maintains a hydrogen bond with the IMPα2 Asn 235 residue of the P1 site ( Figure 4B

Biochemical Assay Demonstrates AAV Po1 VP1 Has Specificity for IMPα Interaction over IMPβ Indicating Use of Classical Nuclear Import Pathway
EMSAs were utilized to assess AAV Po1 Cap-BR interactions with importin isoforms, including IMPβ. Through three independent experiments, it was determined that AAV Po1 Cap-BR could bind with IMPα across the three subfamilies SF1 (α1, α2); SF2 (α3); and SF3 (α5, α7), but did not bind with IMPβ ( Figure 5). These data show an obvious preference for binding IMPα, consistent with a classical nuclear import pathway whereby the NLS binds to the IMPα adapter, which in turn binds IMPβ to form a complex prior to translocating to the nucleus. A lack of binding to IMPβ ( Figure 5) suggests that IMPβ cannot act as a direct binder with AAV Po1 VP1 for non-classical localization, but rather needs IMPα for translocation into the nucleus. box) in sequence order. Binding interactions are displayed through hydrogen bonds (blue dotted lines) and salt bridge interactions (red dotted lines) as indicated by the PDBsum server.

Biochemical Assay Demonstrates AAV Po1 VP1 Has Specificity for IMPα Interaction over IMPβ Indicating Use of Classical Nuclear Import Pathway
EMSAs were utilized to assess AAV Po1 Cap-BR interactions with importin isoforms, including IMPβ. Through three independent experiments, it was determined that AAV Po1 Cap-BR could bind with IMPα across the three subfamilies SF1 (α1, α2); SF2 (α3); and SF3 (α5, α7), but did not bind with IMPβ ( Figure 5). These data show an obvious preference for binding IMPα, consistent with a classical nuclear import pathway whereby the NLS binds to the IMPα adapter, which in turn binds IMPβ to form a complex prior to translocating to the nucleus. A lack of binding to IMPβ ( Figure 5) suggests that IMPβ cannot act as a direct binder with AAV Po1 VP1 for non-classical localization, but rather needs IMPα for translocation into the nucleus.

Biochemical Assessment of Binding Affinities Supports the Model of an AAV Po1 VP1 Bipartite NLS
FP assays were undertaken to not only determine qualitatively the capacity for AAV Po1 Cap-BR to be bound by different importin isoforms (including IMPβ), but also to quantitatively measure the differences in binding affinity between AAV Po1 VP1 monopartite NLS [47] and the bipartite NLS revealed in this study. We observed a clear difference in the strength of binding between monopartite and bipartite NLSs ( Figure 6). AAV Po1 Cap bipartite NLS showed a very strong binding affinity for all importin-α isoforms, with binding occurring in the low nM range, but only very weak binding with IMPβ ( Figure 6B). In contrast, the monopartite AAV Po1 Cap-NLS showed severely reduced binding with all IMPα isoforms compared to AAV Po1 Cap-BR, and no detectable binding to IMPβ ( Figure 6A).

Biochemical Assessment of Binding Affinities Supports the Model of an AAV Po1 VP1 Bipartite NLS
FP assays were undertaken to not only determine qualitatively the capacity for AAV Po1 Cap-BR to be bound by different importin isoforms (including IMPβ), but also to quantitatively measure the differences in binding affinity between AAV Po1 VP1 monopartite NLS [47] and the bipartite NLS revealed in this study. We observed a clear difference in the strength of binding between monopartite and bipartite NLSs ( Figure 6). AAV Po1 Cap bipartite NLS showed a very strong binding affinity for all importin-α isoforms, with binding occurring in the low nM range, but only very weak binding with IMPβ ( Figure 6B). In contrast, the monopartite AAV Po1 Cap-NLS showed severely reduced binding with all IMPα isoforms compared to AAV Po1 Cap-BR, and no detectable binding to IMPβ ( Figure 6A).  1 nM), and IMPα7 (294.2 nM). Binding with IMPβ was so low that an accurate K D was not able to be calculated. (B) FP assay measuring the direct binding between the bipartite AAV Po1 Cap-BR and respective importin isoforms. Strong binding in the 45 pM-1 nM range was observed with IMPα1 (382.3 pM), IMPα2 (1.001 nM), IMPα3 (45.5 pM), IMPα5 (139.6 pM), and IMPα7 (290.3 pM). Much weaker binding was observed with IMPβ, and the binding affinity was difficult to calculate as saturation was not obtained. Error bars for panels A and B were calculated using the standard deviation from the mean of 3 independent experiments. (C) K D of direct binding between importin isoforms and AAV Po1 Cap peptides in nM concentration. K D difficult to calculate due to low binding is represented by N/A. The error was the standard error of the mean.

Discussion
Here we describe the structure of AAV Po1 Cap-BR bound to IMPα, identifying a bipartite binding mechanism. This bipartite binding greatly increases the AAV Po1 VP1 affinity for binding host IMPα proteins when compared to the monopartite AAV Po1 Cap-NLS [47]. This bipartite NLS extends from the common BR1 observed across almost all AAVs to a BR2 further downstream, rather than the typical BR2 found on other AAVs. Due to the manner in which the capsid proteins are translated, BR1 and BR2 are present on VP1 whilst BR1 is not found on VP2. This is of particular note considering that the N-terminus (and its cohort of basic regions) of VP1 is suspected of driving the virion into the nucleus due to exposure through the virion pore [42,68,69]. Our data support this notion, considering the overall difference in the strength of binding found between monopartite AAV Po1 Cap-NLS, bipartite AAV Po1 Cap-BR, and respective IMPα isoforms. Considering VP2 would only contain the weaker binding monopartite NLS (BR2) whilst VP1 holds the bipartite NLS (BR1 and BR2), it can be reasoned that VP1 would be more efficient in mediating nuclear localization. VP3 lacks all basic regions, and in the absence of VP1 and/or 2, does not accumulate in the nucleus [70,71].
BR1 is common across all AAV VP1 proteins including those lacking the typically expected BR2 and BR3. This information paired with the evidence that the inclusion of AAV Po1 VP1 BR1 in the bipartite NLS leads to stronger binding with IMPα than without, gives valuable insight into the way in which BR1 of other AAVs may be either assisting in driving VP1 nuclear localization, or even possibly be driving binding with importin proteins.
Whether weaker interactions between IMPα and the VP2 monopartite NLS also assist the function of VP1 and contribute to nuclear translocation of the virion remains to be directly investigated. Future nuclear translocation studies should be performed using recombinant AAV virions, wherein VP1 possesses a normal BR1 and BR2, whilst VP2 has a mutated BR2 to determine if this monopartite VP2 NLS contributes to transduction at all.